hello,everyone,
i am designing a bandgap reference on the HHNEC process which offer me three kinds of resistros,rdiff,rnwell and rpoly.my topology of BGR is a classical one(PTAT+bipolar).the circuit generate 1.245V when rpoly is used,but 1.27V when rnwell is used,i think 1.245V is reasonable ,but why is it 1.27V when rnwell is used ?
PS:i also get 1.245V when i use ideal resistor to simulate
As I see the reason is different TC dependences of resistors. Unfortunately, the well resistor has a temperature dependence approximately given by R(T)= R0*T^n/T0^n, where n = 2.2. The error caused by this temperature dependence leads to a bit higher reference voltage (Vref0 under T0).
Surely the temperature coefficient of the resistors is largely irrelevant as the ratios will track, which is what is important? As the resistors will have different voltages the voltage dependency is not identical.
thanks for your reply
maybe konstanu is right for that the voltage reference drop to 1.246V after i substitute the tc1 of rnwell with that of rpoly.(tc1 of rnwell is 3e-3;and tc1 of rpoly is -3e-4)
konstanu,could you tell me about more about R(T) and its relationship with voltage reference,or recommand me some meterial?because i never consider res's temperature coefficient,i just check their ratio
That isn't logical. All you are doing is modifying the nominal value by using the wrong temperature coefficient. Also, you seem to be only talking about the nominal value of the bandgap not the value versus temperature which usually gives a better idea of what is going wrong.
hello,heith,thanks for replying
i didnot want to modify anything,i just want to know where the difference(1.245V and 1.27V) come from,since the voltage reference should be fixed based on the formula Vref=Vbe+(1+R2/R1)*deltaVbe.
I would suggest you look at the voltage versus temperature in each case. Also, check the actual values of the resistors in each case (probing current and voltage rather than the value/dimensions you set).